What happens to gestures when Italians do not speak Italian? A study of gestures on native speakers of Italian

Radboud University

Faculty of Arts

What happens to gestures when Italians do not speak Italian?

A study of gestures on native speakers of Italian

Master Thesis

Language and Communication Coaching

Student name: Sara Cerulli

Student number: 1011853

Supervisor: Jarret Geenen

Table of content

Speech and gesture transcriptions ... 26

Analysis and results ... 29

Quantitative analysis ... 29 Qualitative analysis ... 31 Discussion ... 55 Conclusion ... 67 References ... 69 Appendices ... 73

English transcription (higher proficiency speakers) ... 73

English transcription (lower proficiency speakers) ... 75

Italian transcription (higher proficiency speakers) ... 77

Italian transcription (lower proficiency speakers) ... 79

English literal translation of Italian transcription (higher proficiency speakers) ... 81

English literal translation of Italian transcription (lower proficiency speakers) ... 83

English idiomatic translation of Italian transcription (higher proficiency speakers) ... 85

English idiomatic translation of Italian transcription (lower proficiency speakers) ... 87

Transcript of gestures (higher proficiency speakers) ... 91

Transcript of gestures (lower proficiency speakers) ... 100

i

Acknowledgments

I would like to thank my supervisor, which has guided and supported me throughout the thesis writing process. His knowledge and his insight were of fundamental importance to me. Moreover, this project would not have been possible without the people that helped me in carrying out the experiment, the participants, and most importantly the listeners: Matilde, Marsela, Maria, Natalia, and Alan. Last but not least, a big thank you goes to my family, to my boyfriend, and to my friends who encouraged me and pushed me to do better, always.

1

Abstract

Speech is often accompanied by gesticulation. In communication, people often employ not only the verbal channel, but also the gestural one. Gesture and language are strongly interconnected and co-expressive, thus it has been posited that they should be regarded as two aspects of a single process. By employing the method of storytelling, the current study investigated speech-gesture production of Italian native speakers their first and second language, English. It has been suggested that the use of co-speech gesture might be different in a bilingual’s L1 and L2. Some have found an increase in gesture frequency in bilinguals’ weaker language, while others have reported that gesture frequency is higher in bilinguals’ L1. In the present study, gesture frequency was slightly higher in the L2, especially for lower proficiency speakers. Moreover, in the narrations similar iconic gestures were performed by the same speaker when narrating the same portion of the story in the two languages. However, iconic gestures produced by lower proficiency speakers in English displayed higher degrees of iconicity compared to the coupled Italian gesture. Theories and implications are discussed.

2

Introduction

Natural speech is often accompanied by gesture. People communicate not only through speech but also through manual gestures. Their role in communication has become so important that research on this phenomenon is abundant, also thanks to the pioneering work of McNeill (1985) and Kendon (1997). The researchers argue that gesture is a fundamental component of everyday communication and is significant as speech itself. Speech and gesture are tightly linked, and since they are produced together, they should be regarded as two aspects of a single process. Explanations with regards to the underlying cognitive processes that link speech and gesture differ in the literature. One line of thought sees speech as primary and gesture as auxiliary. Gestures are seen as facilitating lexical retrieval (the Lexical Retrieval Hypothesis, Krauss, Chen, & Gottesman, 2000), or as part of the conceptual planning of the message (the

Information Packaging Hypothesis, Alibali, Kita, & Young, 2000). Another line of thought

regards gesture and language as equals, where gesture is itself an integral part of an utterance (the Growth Point Theory, McNeill, 1992; the Interface Hypothesis, Kita & Özyürek, 2003).

Studies on gesture have investigated the phenomenon from different perspectives, and linguistic and cultural differences (or similarities) in gesture production has been an area of research widely examined. Cross-cultural and cross-linguistic research has focused on some aspects of gesture such as frequency variation, spatio-temporal characteristics of gestures, linkage to the language spoken, types of gesture performed. The seminal work conducted by Efron (1941) found cultural differences in gesture production of Italians and Eastern European Jewish native speakers, but also that cultural and social forces might have an effect on communicative norms and non-verbal behavior, including gesture. Apart from the conventional folklore about gesture frequency and prevalence when Italian people speak, several studies, including Efron’s, found the use of gesture by Italians to be prominent. Research conducted by Graham and Argyle (1975), for example, appear to suggest that Italian speakers might rely more heavily on gestures when communicating and comprehending information compared to English speakers. Cross-linguistic variation has also led to substantial research on gestures of bilinguals. As some studies suggest (Marcos, 1979; Gullberg, 1998; Seto, 2000; Nicoladis, Pika, Yin & Marentette, 2007), the use of co-speech gesture might be different in a bilingual’s L1 and L2. Results, however, are conflicting and there is still disagreement. Some have found an increase in gesture frequency in bilinguals’ weaker language, suggesting that language proficiency

3

might be involved in gesture production. Others have reported that gesture frequency is higher in bilinguals’ L1, hinting that frequency might indeed be associated with specific languages.

The present research will examine Italian native speakers performing a narrative task in their first and second language, English. Based on the findings that gesture frequency of Italians is prominent in communication, coupled with the idea that bilingual speakers’ use of gesture might be different in their L1 and L2, it would be significant to understand gesture production frequency across the Italian and the English language while spoken by the same speaker. Gestures do not have standard forms and different speakers may convey the same meanings in idiosyncratic ways (Goldin-Meadow & McNeill, 2012). Thus, examining the same speakers speaking multiple languages allows to account for idiosyncratic variation during the repetition task. This approach may provide further insights into the gesture-language relationship and add to our understanding of how social and cultural forces might affect gesticulations. Since evidence from previous research shows that proficiency might affect gesture production, participants’ level of proficiency will be taken into account. To investigate gesture frequency of Italian native speakers in their L1 and L2, a quantitative approach will be employed.

The second part of the research will focus on iconic gestures, employing a qualitative approach. Research on iconic gesture is abundant and several theories have been developed on their function. McNeill (1985), for example, proposes that speech and iconic gestures “cooperate to present a single cognitive representation” (McNeill, 1985, p. 353). Butterworth and Hadar (1989) posit that gestures have a functional role in word retrieval. Extensive research has been conducted testing both theories, however, the vast majority of research has focused on L1 gesture production. Cross-linguistic research has mainly been conducted on iconic gesture encoding meaning components of motion (e.g. Özyürek, 2002), while, research on L2 learners has mainly focused on frequency variation. This fails to provide an understanding about other potential cross-linguistic differences, which may not be related to frequency or motion encoding. Qualitatively looking at iconic gestures in bilinguals will add to our understanding of such gestures and their function. If, as posited by McNeill and Kendon, speech and gestures are two integrated systems, then gesture becomes important too in the study of languages and may be able to give us insights on L2 speakers and language development. Again, the proficiency level of speakers will be taken into account, as it might be that proficiency influences gesture production.

4

RQ1: Is there a difference in gesture production frequency between English and Italian narrations and does proficiency play a role in frequency distribution across languages?

RQ2: Is there a difference in iconic gesture production across English and Italian narrations

and is this difference contingent on proficiency level?

RQ3: If a difference is found between iconic gestures used in Italian and English, where

does this difference lay?

In the attempt to motivate naturalistic speech-gesture production, a narrative task will be designed, since gestures are known to be abundant in storytelling (Gullberg, de Bot & Volterra, 2008). Italian native speakers will watch an animated cartoon of Tweety and Sylvester and will narrate its story to a listener, who will pretend to be unaware of the plot of the cartoon. The experiment will take place in separate days, whereby one time the language of narration will be English and the other Italian. For the quantitative analysis, gesture will be counted, and the length of the narrations will be taken into account. The ratio of gesture per word will allow to make comparison between speakers’ L1 and L2. To get insights on the potential role of proficiency, participants will be then subdivided into higher and lower proficiency and gestures produced in English by the two groups will be compared, to see if proficiency has a bearing on frequency. For the qualitative analysis, the concept of iconicity will be employed as a criterion to study each gesture, by considering the parameters that are iconic in a gesture and that resemble aspects of the referent, paying specific attention to temporal and rhythmic components of speech-gesture production, and to the phases of gestures that co-occur with speech.

5

Literature review

In recent years, much attention has been placed on the role of gesture in interaction. McNeill (1985) and Kendon (1997) argue that gesture is a fundamental component of everyday communication and is as significant as speech itself. Speech and gesture are tightly linked, to the point that some believe they should be regarded as two aspects of a single process (Kendon, 1997; McNeill, 1985). The pivotal point behind all theories on gesture is the semantic-pragmatic and temporal coordination between speech and gesture. The two modes convey closely related meanings, but they do so in their own ways. Gestures are co-expressive, they contribute to the overall propositional content of the utterance, but they are not semantically redundant. Imagine someone saying “I found a ball” and accompanying speech with a gesture depicting a round shape. Speech and gesture refer to the same semantic content, however, the gesture does not merely reproduce what is already conveyed through speech. The size of the ball illustrated by the hands may help discern, for example, a basketball from a tennis ball. Moreover, speech and gesture are synchronized, in the sense that gestures happen in rhythmic integration with the flow of speech production (McNeill, 2007; de Ruiter, 2007; Kendon, 2004).

The word gesture, however, is imprecise and comprises different phenomena. The interest here is in spontaneous gesticulation (from now on ‘gesture’) used in co-occurrence with speech, thus driven by current meaning and not constrained by conventions. These gestures are incomplete without speech accompaniment. Other types of gesture, such as emblems, pantomime, sign language, still part of communication, relate to speech in different ways (McNeill, 2002; McNeill et al., 2008). Therefore, such gestures will not be discussed in detail in the present research. Emblems are culturally specified, and their interpretation is consistent within a culture, but may differ between cultures (Cassel, McNeill & McCullough, 1999). Pantomime is gesture without speech (McNeill et al., 2008).

In the attempt to classify and categorize gestures, many systems and taxonomies have been created in the past years that vary in the categories gesture may fall into. The one adopted in this thesis is the description proposed by McNeill (2007). He defines four types of gestures and later illustrates how it is better to refer to them as dimensions. The reason for that is because dimensions can be multiple, whereas categories have to exist separately. Gesture is global and synthetic. Global refers to the fact that its meaning is determined by the meaning of the whole. Synthetic means that distinct meanings of a single gesture are concentrated into one symbolic form (McNeill et al., 2008). Therefore, since a gesture can entail multiple meanings, it is not

6

merely iconic, or metaphoric, or deictic. It can be all of the above at the same time. Using dimensions obviates the need for a gesture to fall under a single heading, rather, it can belong to two or more dimensions, without the need to define a hierarchy between them. A taxonomic classification of gesture would undermine the fluid and variable nature of the dimensions embedded in gesture. The distinction proposed by McNeill is the following:

1. iconic or iconicity: those gestures that represent concrete entities or actions. Form and manner of gestures present picturable aspects of the semantic content; for example, if someone raises the hand towards the mouth to imitate the act of eating;

2. metaphoric or metaphoricity: those gestures that represent images of the abstract, in which an abstract meaning is presented as form or space; an example would be a circling movement of the hand to indicate “time passing”;

3. deictic or deixis: those gestures used to point. Gesture can embody concrete deixis (locating entities and actions in space) and abstract deixis (space is used to present non-spatial meaning);

4. beat or temporal highlighting: movements of the hands that seem to ‘beat’ time along with the rhythm of speech. Such gestures may be used by the speaker to signal the temporal locus of something s/he feels is important in speech.

It is abundantly clear that gestures are naturally attended to during the communicative process (Cassel et al., 1999) and that they carry semantic information which is relevant and incorporated by interlocutors during interaction (Beattie & Shovelton, 1999a, b; Cassel et al., 1999). There is general agreement that gesture is temporally integrated with speech production (Kendon, 2004; McNeill, 2007), can assist in language production (Kita, 2000) and is semantically and pragmatically inseparable from speech output in face to face interaction (McNeill, 1985; Kendon, 2004). While ample evidence supports the above propositions, multiple questions remain regarding the cognitive integration and precise relationship between language and gesture. One significant domain of research has sought to elucidate the extent to which gestures are associated with specific languages (Efron, 1941; Graham & Argyle, 1975; Kita & Özyürek, 2003; Pika, Nicoladis & Marentette, 2006). Given the relationship of gesture and language during speech production, much of the work in this domain hopes to tease apart the precise relationship of language and speech cognitively. Central here is the extent to which gestures are or are not tied to or related to the specific language being spoken at any given time. Kita and Özyürek’s (2003) gesture production model and their multilingual research on gesture encoding during speech production suggests that the types of gestures produced during

7

speech have a close relationship to the specific language being spoken insofar as particular semantic features from the affordances of the language affect the encoding of the gestures themselves. Their Interface Hypothesis was based on findings of a study where they examined the cognitive process of gesture production, and how the information coordination between co-speech gesture and co-speech is achieved. Three different theories were compared: The Free Imagery Hypothesis (de Ruiter, 2000; Krauss, Chen, & Chawla, 1996); the Lexical Semantic Hypothesis (Butterworth & Hadar, 1989), and the Interface Hypothesis. According to the first, gestures are generated pre-linguistically, from imagery in the working memory. Therefore, they are not influenced by the representational potential of the language. The second hypothesis maintains that gestures are generated from the semantics of the lexicon of the concurrent speech. Thus, information that is not verbalized will not be encoded in gestures. The third hypothesis argues that gestures originate from an interface representation between speaking and spatial thinking. Gestures both encode non-linguistic properties and structure the information in a way to make it relatively compatible with linguistic encoding possibilities.

The authors analyzed gestures produced in narratives by native speakers of English, Turkish, and Japanese. The participants watched an animated cartoon and were asked to subsequently narrate the story to a listener who did not know the plot. The foci of the research were two: the effect of limited linguistic resources on gestural representation and the effect of different clausal packaging of spatial information on gestural representation. First, it was found that Turkish and Japanese represented the event in question without the feature difficult to verbalize in their language, and this was reflected in the gestural representation. Second, since the information to be expressed is linguistically packaged more concisely in English than in the other two languages, as opposed to English, Japanese and Turkish spread the information into more processing units for speech production (approximately a clause). As a result, separate gestures were used to express the information. Moreover, in both instances, information that was not expressed linguistically was encoded in gestures. The results support the Interface Hypothesis which predicts that gestures are simultaneously shaped by the linguistic packaging of the relevant information and by the spatio-motoric features of the referent that are not verbalized.

While Kita and Özyürek’s findings suggest that the language being spoken demonstrably influences gestural encoding, the nature of the stimuli, which were only two motion events, makes it difficult to generalize about the precise influences a language may have on gesture production in instances of natural discourse. The method also restricts our

8

understanding of cross-linguistic influences to a single gesture type - namely an iconic gesture encoding manner and path. Additionally, their focus on semantic encoding limits the ability to argue much of substance about the nature of the relationship between specific languages and gesture. While semantic feature distribution is clear, gestures do not only serve semantics. Gestures have deictic and pragmatic functions as well, which vary cross-culturally, and gestures are also produced at different rates across cultures. The study provides little insight into the potential of frequency variation. Given the independent sample design, which was necessary for their particular task design, the findings fail to account for idiosyncratic variation across cultural groups during the repetition task. This last shortcoming is especially problematic. All speakers of a particular language do not gesture in a specific way. As evidence in the results section of this thesis, it appears as though there may even be a gestural idiolect to some extent. In other words, speakers of multiple languages may indeed use similar gesture repertoires across languages, and this puts the idiosyncrasy of the gesture-specific language relationship into question. However, this gesture-specific language relationship and how gesture may be influenced by socialization and culture is a nuanced subject and contradictory findings suggest different relationships.

Historically, research has actually suggested that there is a large influence of social and cultural forces which may affect gesticulations (Efron, 1941). Thus, cross-cultural variation in gesture has led to substantial research in efforts to understand the gesture-speech relationships. One particular culture which has figured significantly are Italians. Apart from the conventional folklore about gesture frequency and prevalence when Italian people speak, there is historical credence to the significance of gestures in Italy given the repertoire of gestures used by Italians. In the nineteenth century the first collection of Italian gestures was created by the Canon de Jorio (1832). More recently, Munari (2000) wrote a book on Italian gesticulation and non-verbal communication. It is a real dictionary of gestures, written in four languages, in which the author illustrates the history of Italian gestures, many of which were Neapolitans and then became known nationally and some even worldwide. However, it is not just folklore, general interests, or taxonomies which attest to the interest in Italian gestures. Considerable empirical work has investigated gestures in relation to specific cultures.

Efron (1941) was one of the first who showed that there may be gesture variation across cultures. His research investigated the influence of race and environment upon behavior and found that, rather than race, it is the cultural environment in which people live that influences non-verbal behavior. The subjects of the study were both traditional and assimilated Southern

9

Italian and Eastern European Jewish immigrants in the United States. Efron compared gestures produced by traditional Italians and traditional Eastern European Jewish and found that there were indeed cultural differences in the use of gestures produced by the two groups. Particularly, such differences were visible in spatio-temporal aspects, such as the incidence and frequency of gesture production, and in the linguistic functions of those gestures. Gestures produced by traditional Italians were abundant and the author discovered an extensive vocabulary of gestures that was widely shared. Such representations were found to be consistent and elaborate. However, it was also found that the gestures of the Italians and Jewish assimilated to the American culture differed far less from each other and, interestingly, resembled those of the American group with which they had become associated. What the findings seem to suggest is that despite heritage, cultural and social forces may have a considerable effect on communicative norms, and thus may influence non-verbal behavior. As a consequence, it might be that the use of gesture by bilinguals might be different from that of monolinguals, depending on the cultural environment in which they live and the cultural group they are associated with. Efron’s study was not the only one that reported differences in the amount of gesture used by Italians in comparison to other cultures. When investigating motion events described by speakers of two typologically different languages, Danish and Italian, Wessel-Tolvig and Paggio (2016) found that, when describing a motion event, Italian speakers tended to produce gestures twice as often as Danish speakers. Moreover, high frequency of gestures performed by Italians compared to other cultures was also visible in children. Iverson, Capirci, Volterra and Goldin-Meadow (2008) conducted a study on American and Italian children and discovered differences in the gesture repertoire of the two groups. Italian children produced more representational gestures than the American children. Moreover, most of them were non-redundant and included object, action, and attribute gestures. Despite the small sample size, it is likely that the findings of the study may reflect differences in the nature of the gesture models to which children are exposed and the cultural environment in which they live, as was also predicted by Efron (1941).

Another study on gesticulation and Italians and on the cultural differences of Italian and English gestures was conducted by Graham and Argyle (1975). The authors investigated the role of gestures in completing the meaning of utterances with the aim to discover whether L1 English speakers and L1 Italian speakers differed in their use of gestures. The authors indicate how culture plays a role in gesture performance. Italian is defined a high gesture culture in terms of frequency, whereas the use of gesture is less frequent in English, termed a low gesture

10

culture. However, it is not clear whether Italians communicate additional information by doing so. With this in mind, Graham and Argyle hypothesized that the Italians would benefit more than the English from the use of gestures, which would translate into an increase in the amount of information conveyed when gestures are allowed for Italians.

British English and Italian students were asked to describe pictures containing shapes of high and low verbal codability to speakers of the same language. In one condition speakers were allowed to use gestures to describe the pictures, in the other condition gestures were not allowed. In both conditions, listeners were asked to draw the shape that was being described. These drawings were scored and analyzed for similarity to the original. The results showed that when gestures were allowed performance improved and both Italians and English were more accurate in communicating shapes. What is primarily significant is that for the Italians the percentage of improvement in performance when gestures were allowed was greater than for the English. This provides some evidence that frequency may additionally serve a communicative function, indeed, when gestures were allowed, Italians actually communicated more information than the English. Therefore, it would appear that L1 Italian speakers might rely more heavily on gestures when communicating and comprehending information that do L1 English speakers. The nature of the task, however, which required participants to describe shapes of high and low verbal codability, is not very informative about the behavior of the two cultural groups in instances of natural communication. Therefore, further research would be necessary to establish whether the findings can be generalized to natural discourse in interaction, when it is not material of high/low verbal codability that is being communicated.

As Efron (1941) results seem to suggest, cultural and social factors might influence non-verbal behavior and, therefore, gestures produced by bilinguals might differ from that of monolinguals. Abundant research has investigated bilinguals’ use of gestures in their L1 and L2. An area of focus has been the quantitative differences in gesture production of bilinguals between L1 and L2 (Marcos, 1979; Gullberg, 1998; Seto, 2000; Nicoladis et al., 2007). However, the results of this work are conflicting. Such discrepancy among results might be due to differences across methodologies used or might be caused by other factors. Marcos (1979) and Gullberg (1998) have found an increase in gesture production when speakers used the L2 compared to their L1. Marcos (1979) found that English-Spanish and Spanish-English bilinguals performed more gestures when they were speaking their weaker language and Gullberg (1998), when investigating French and Swedish bilinguals, found that both groups produced more gestures when speaking their L2. This seem to suggest that matters of

11

proficiency are at play when considering gesture frequency across languages. More specifically, that gesture frequency increases during speech production in a weaker language. However, other studies show contrasting results. Seto (2000), for instance, examined Japanese speakers of English and Australian speakers of Japanese. The findings revealed that for both groups the frequency of gestures was higher when participants were speaking in English. In contrast, this seems to suggest that gesture production frequency is not a matter of proficiency and may indeed be linked to a specific language. In essence, whether gesture frequency is associated with specific languages or with proficiency in a language remains somewhat unknown. Furthermore, the theoretical nuances regarding why gesture frequency may be higher in either a language classically recognized as being gesture-rich or one’s weaker language remain preliminary and unsubstantiated. Finally, it could be that, in many cases, conflicting findings may simply be attributable to incongruent methodologies employed across studies.

Research on bilinguals’ use of gesture has also sought to link differences and similarities in gesture production between a speaker’s L1 and L2 to the notion of transfer (Cavicchio & Kita, 2013; Pika et al., 2006). Language transfer, a phenomenon largely examined in studies on language, is the impact that existing languages has on the acquisition and production of a new one (Gullberg, 2014). The notion of gesture transfer, however, remains problematically under clarified. Implicitly, the notion of transfer suggests that frequency, type, and nature of gesticulations are intimately linked with a specific language and can be transfer from one language to another as one switches language. However, if such features are not linked to a particular language, then there is nothing to be transferred. This also implies that at certain proficiency levels, gesture, or at least gestural dispositions, remain unobstructed by second language acquisition. In extension, this results in a particular view of what exactly gesture is – that gesture is associated either with semantic memory, and attempts to lexicalize particular concepts will result in recurrent gestures due to mutual storage, or that gesture is associated with particular lexical affiliates in language production, resulting in lexemes of the same semantic scope being accompanied by gestures previously associated with one’s first language. In support of this, some studies have shown how speakers of an L2 do not necessarily gesture like target-language speakers but show traces of their L1 in their gesture production (Gullberg, 2014). Thus, as mentioned above, there is the possibility that gesture, like language, can be transferred cross-linguistically. However, the evidence for gestural transfer in the literature is mixed. Research by Cavicchio and Kita (2013) investigated the relationship between gesture and language in English/Italian bilinguals. The focus of the research was gesture rate and

12

gesture space. Both aspects are known to vary cross-culturally, and before their study, gesture size in bilinguals had never been investigated. Gesture size was operationalized as how gestures are performed in space, dividing the gesture space in two sectors, center and periphery. 30 participants, of which 10 English monolinguals, 10 Italian monolinguals, and 10 English-Italian bilinguals, took part in the experiment. They watched a 10-scene cartoon and retold it to a listener. Monolinguals told the story twice in their L1, whereas bilinguals told the story once in Italian and once in English.

As previous research found, gesture rate was higher for Italians than for the English. Moreover, gestures produced by Italians were more salient compared to gestures produced by the English. However, when comparing gestures produced by English-Italian bilinguals, no evidence of transfer was found with regards to rate and salience. When bilinguals switched language, their gesture parameters switched accordingly. Bilinguals gestures, however, were overall more salient than those of monolinguals. This could be due to the fact that bilinguals could be weaker in one of the languages and make their gestures more salient in order to facilitate communication. To account for lack of transfer, the authors resorted to La Heij’s (2005) concept selection hypothesis, whereby the selection of some parameters, such as language and gesture, occurs at a pre-linguistic level. Thus, the features specifying language and gesture parameters might be selected at a high-level processing stage, in which verbal and nonverbal aspects of communication are planned together (Cavicchio & Kita, 2013). What is problematic with the study, however, is that the authors considered the English-Italian bilingual group as a whole, despite mentioning that some were English and some Italians. This conflation of the linguistic demographics of the sample creates considerable issues in validity. In the research it was not specified which language bilinguals acquired first, whether they acquired both languages at the same time, or in which country the bilingual participants grew up and lived. As we know from Efron’s research, the communicative culture may have a considerable effect on communicative norms of people. Failing to make such distinctions between the heterogeneous group of bilingual participants may lead to potential differences within the group not being detected.Moreover, no explanation was given as to why monolinguals performed the task twice, which of the two storytellings was chosen for the analysis, and what were the criteria for choosing one narration instead of the other. This might have had consequences for the results of the analysis conducted.

Another study on gesture transfer, which presented opposite results, was conducted by Pika et al. (2006), who investigated gesture frequency cross-linguistically and the possibility of

13

gestural transfer occurring from a high frequency gesture language to a low frequency gesture language. Since evidence of linguistic transfer has been found in the literature, adopting McNeill’s (1985) notion that speech and gesture form a single integrated system, then it is possible that gestural transfer might occur as well. The authors compared French–English bilinguals, English–Spanish bilinguals, and English monolinguals. They participated in a storytelling task whereby, after watching a cartoon, speakers had to narrate the story to a listener. Bilinguals told the story twice, once in their L1 and once in their L2, whereas monolinguals told the story once. The analysis focused on gesture rate and frequency of production of gesture types.

The results demonstrated that gestural transfer occurred from a high- to a low-frequency gesture language, showing that the overall gesture rate of French–English bilinguals and English–Spanish bilinguals was higher than the English monolinguals. This was particularly prominent for iconic gestures. Moreover, the study showed that gestural transfer can also occur from an L2 to the L1. Bilinguals whose L1 was English gestured more frequently in English than English monolinguals. Therefore, the researchers maintain, for transfer to occur it does not matter whether the high frequency gesture language is the L1 or the L2. Higher gesture rate in both bilingual groups compared to the English monolinguals was the factor that allowed to determine the occurrence of transfer from a high to a low frequency gesture language. The authors maintained that “The only way that second language learners could know that a language was a high frequency gesture language was through exposure to multiple native speakers of that language” (Pika et al., 2006, p. 324). As a matter of fact, all bilingual participants who took part in the experiment spent at least one year in a country where their L2 was spoken. However, to state with certainty that exposure to the L2 culture contributed to gesture transfer, this factor should have been controlled for. Testing two more groups, English-Spanish and French-English bilinguals who never lived in a country were their L2 was spoken, is necessary to confirm their claim. What also raises some questions is that, when testing French-English bilinguals’ gesture frequency in the two languages, it was found that participants produced significantly more gestures in English than in French. If gesture transfer did occur from a high frequency gesture language, French, to a low frequency gesture language, English, it actually occurred to a greater extend, even more gestures were produced in the low than in the high frequency gesture language. Transfer, therefore, might not be the only way to account for the results. Including a French and a Spanish monolingual group might have provided stronger evidence for the authors’ assertions. Lastly, the study restricts our

14

understanding to transfer with regards to frequency and types of gestures. It remains to be seen whether other features of gestures can be transferred, and whether transfer can occur from a low to a high frequency gesture language, in the sense that low frequency of gesture is displayed even when a speaker is using a high frequency gesture language, due to influence of a low frequency gesture language.

In summation, studies suggest that Italians frequently accompany their speech with gesture, and gesture more compared to some other cultures. Cultural differences in gesture production is a concept widely agreed upon. Efron’s (1941) seminal work provided some evidence that cultural and social forces may have a considerable effect on communicative norms, and thus may influence non-verbal behavior. In particular, the results of his research showed that Italian native speakers gestured more compared to Eastern European Jewish and had a widely shared vocabulary of gesture. Graham and Argyle’s (1975) results showed that Italians produced more gestures than the English, which resulted in an increased improvement in performance for Italians. This provides some evidence that frequency may additionally serve a communicative function and suggest that Italians might rely more heavily on gestures when communicating and comprehending information that do the English. In Wessel-Tolvig and Paggio’s (2016) study, Italian native speakers produced gestures twice as often as Danish speakers when describing motion events. Lastly, Cavicchio and Kita (2013) reported that gesture rate was higher for Italians than for the English, and gestures produced by Italians were more salient. Based on these findings, coupled with the idea that bilingual speakers’ use of gesture might be different in their L1 and L2, it would be significant to understand gesture production frequency across the Italian and the English language while spoken by the same speaker. Gestures do not have standard forms and different speakers may convey the same meanings in idiosyncratic ways (Goldin-Meadow & McNeill, 2012). It is not that all speakers of a particular language gesture in a specific way and, in this sense, gesture is not inherent to a specific language, it might only be inherent to a speaker of that language. For this reason, to make inferences about the gesture-specific languages relationship, it would be necessary to take idiosyncratic variation across cultural groups into account and thus, investigate the same speakers speaking multiple languages, rather than different speakers speaking different languages.

Some studies have reported that bilinguals gesture more frequently in their L1 (Cavicchio & Kita, 2013; Gregersen, Olivares-Cuhat & Storm, 2009), which seems to suggest that gesture rate is associated with specific languages, while others that bilinguals produce more

15

gestures in their L2 (Marcos, 1979; Gullberg, 1998), which, on the contrary, seems to suggest that proficiency has an influence on gesture rate. However, whether gesture frequency is associated with specific languages or with proficiency in a language it is still unknown. Italian is said to be a high frequency gesture language and culture. English, on the contrary, is said a low frequency gesture language. What would be interesting to study is whether Italian bilinguals display a high rate of gesture even when they are not speaking their native language but their L2 (in the case of the present study English). It might be that Italians will display the same gesture rate across languages, which may allow to determine the occurrence of transfer from a high to a low frequency gesture language, but also that gesture rate is similar in the two languages since speakers of multiple languages may use similar gesture repertoires across languages, or it could be that proficiency might affect gesture production, which would result in a higher gesture rate in participants’ L2. Lastly, if gesture rate is found to be higher in Italian, it might be that participants switch parameters and thus, gesticulate less as they speak a low frequency gesture language, or it might be that frequency is inherent to the Italian language and that when Italian native speakers switch language their gesture attitude is lost. Therefore, the first point of the research will focus on gesture production frequency in Italian and English while spoken by Italian native speakers. Since evidence show that proficiency might affect gesture production, participants’ level of proficiency will be measured systematically. To motivate naturalistic speech-gesture production a storytelling task will be employed. Participants will watch a short cartoon and narrate the story to a listener in English and in Italian. The aim is to discover whether Italian-English bilinguals switch gesture parameters when they speak their L2, and thus gesticulate more or less, or whether gesture production frequency is kept constant regardless of whether Italians are speaking their L1 or L2. The current study proposes the following research question:

RQ1: Is there a difference in gesture production frequency between English and Italian

narrations and does proficiency play a role in frequency distribution across languages? Research has shown that gestures produced by bilinguals in their L1 and L2 not only differ in terms of frequency of use but might also differ in terms of types (or dimensions) of gestures used. Nicoladis et al. (2007), for example, found that more iconic gestures were used in participants’ L2. Moreover, there is evidence suggesting that the production of certain types of gestures might be influenced by proficiency in a language. Research by Gregersen et al. (2009) found that advanced learners of Spanish tended to produce more iconic gestures compared to intermediate and beginning learners. So, Kita and Goldin-Meadow (2013) found

16

that proficient speakers tended to produce iconic gestures to further specify referents already specified in speech, and concrete deictic gestures for referents that were not specified in speech, whereas less proficient speakers tended to produce both iconic and concrete deictic gestures regardless of referents being lexically specified in speech. Extensive research has been conducted on the role and function of iconic gestures. However, the vast majority of research has focused on iconic gestures produced by native speakers. Studies comparing iconic gestures in different languages has mainly been conducted on iconic gesture encoding meaning components of motion, like path and manner produced by speakers of typologically different languages (e.g. Özyürek, 2002). Moreover, research on second language learners has mainly focused on frequency variation across the L1 and the L2. This fails to provide an understanding about other potential cross-linguistic differences, which may not be related to frequency or motion encoding processes. Qualitatively looking at iconic gestures in a speaker’s L1 and L2 might help us to add to our understanding of such gestures and their function. Moreover, if we consider speech and gestures to be two integrated systems, as posited by McNeill (1985) and Kendon (1997), then gestures become important too in the study of languages and may be able to give us insights on L2 speakers and language development. As Gullberg et al. (2008) argue, analyzing gestures and speech together may provide a fuller picture of learners’ strategies of problem-solving. In addition, taking proficiency into account might help identify and discern potential communicative strategies employed by L2 speakers that have different proficiency levels. The research will employ a qualitative approach to investigate the following questions:

RQ2: Is there a difference in iconic gesture production across English and Italian narrations

and is this difference contingent on proficiency level?

RQ3: If a difference is found between iconic gestures used in Italian and English, where

does this difference lay?

Two major contrasting theories have been developed on the function of iconic gestures, one proposed by McNeill (1985) and the other by Butterworth and Hadar (1989). According to the first, speech and iconic gestures “cooperate to present a single cognitive representation” (McNeill, 1985, p. 353). McNeill presents a number of examples of speech-gesture co-occurrence, the following displaying how this speech-gesture cooperation occurs:

‘she chases him out again’ [hand, gripping an object, swings from left to right]

17

Here meaning is conveyed through both channels. However, the iconic gesture in question “conveys the idea of the instrument of the act, [an umbrella,] whereas the act itself is described in the concurrent sentence. [. . .] To get the full cognitive representation that the speaker had in mind, both the sentence and the gesture must be taken in account” (McNeill, 1985, p. 353). McNeill’s central argument, thus, is that iconic gestures act in cooperation with speech to convey ideas. Butterworth and Hadar (1989), however, have a quite different view on iconic gestures. They suggest that such gestures have a functional role in word retrieval. Their theory follows Butterworth and Beattie’s (1978) work whereby it was found that iconic gestures “tended to have their onsets in pauses in relatively fluent sections of speech, and hence preceded the onset of related speech material and offer an account in terms of the relative difficulty in accessing the intended lexical items” (Butterworth & Hadar, 1989, p. 170). Failure to retrieve the phonological form of a lexical item would delay speech output while unaffecting iconic gesture onset, the latter revealing this lexical access difficulty. The following example comes from Beattie and Aboudan’s (1994) study where respondents narrated a cartoon story:

‘(pause) starting it at the front with the (pause) winder thing’ [hand moves in a winding movement]

In this example, the iconic gesture starts and ends in between two pauses, before the lexical affiliate is even uttered. The speaker, unable to access the lexical item “starter-handle” employed “the winder thing”. This example seems to fit better into Butterworth and Hadar’s theory whereby iconic gesture might have a functional role in word retrieval.

Extensive research has been conducted testing both theories. Beattie and Goughlan (1999) tested Butterworth and Hadar’s (1989) theory experimentally inducing the tip-of-the-tongue (TOT) state in participants. This is a particular type of lexical accessing problem that occurs when “We are sure that the information is in memory but are temporarily unable to access it” (Brown, 1991, p. 204). Participants in the experiment were presented with a list of 25 definitions of words from which they were asked to recall the target words. Half of the participants were instructed to fold their arms in order to prevent gesturing. From Butterworth and Hadar’s theory it was predicted that more correct words should be recalled by participants who were free to gesture, and in a TOT state, they should resolve more TOTs than those who had their arms folded. Moreover, they predicted that the TOTs should be associated with the production of iconic gestures, and that the latter should be involved more in resolved TOT states than unresolved ones.

18

Results, however, did not completely support Butterworth and Hadar’s theory that iconic gestures have a functional role in word retrieval. Contrary to prediction, more words were recalled by participants who had their arms folded, although when TOTs occurred, those who were free to gesture resolved more TOT states. Despite gestures being present in TOT states, iconic gestures were the least employed in word retrieval. A further analysis examined whether gestures were associated with TOTs being resolved. Surprisingly, results revealed that significantly more TOT states were resolved when gestures were absent than when they were present. Overall, despite showing that gestures in general are associated with lexical search, the results of Beattie and Goughlan’s (1999) study failed to find real evidence for Butterworth and Hadar’s theory.

Beattie and Shovelton (1999a, b) and Holler and Beattie (2003) have conducted a series of studies to test McNeill’s theory that iconic gestures accompanying speech convey critical information in interpersonal communication. The early studies (Beattie & Shovelton, 1999a, b) examined information conveyed by gestures that were presented to participants on video. A later study, however, also examined the phenomenon in conversational interaction (Holler & Beattie, 2003). In Bettie and Shovelton (1999b) 14 participants were asked to narrate cartoon stories that were videotaped. Iconic gestures were selected from the narrations. 10 other participants responded to a structured interview about information contained in the video narrations in three conditions: video, audio-only, and vision-only. The questionnaire asked questions about semantic categories (identity, number, description of action, shape, size, movement, direction, speed, relative position) to ascertain what information the respondents picked up from the clips. The interviews were then analyzed and compared to the original cartoons.

From the results, it appears that iconic gestures convey additional meaning. With the video presentation, respondents were significantly better at answering questions about the semantic properties of the original cartoons than in the audio-only presentation. Specifically, this was particularly relevant for two semantic categories: size and relative position. These seem to be the semantic properties more accurately encoded in iconic gesture across the sample. Moreover, it was found that “iconic gestures in the absence of speech also communicate significant amounts of information about the world, in the sense of transmitting significant amounts of information about it” (Beattie & Shovelton, 1999b, p. 453). In fact, the overall accuracy in the vision-only condition was 20.4%. Thus, the study showed not only that iconic gestures communicated additional information, and that they did so even in absence of the

19

accompanying speech, but also what semantic categories were conveyed through gestures with regards to narration of cartoon stories.

One of the shortcomings of Bettie and Shovelton’s (1999b) study is that information conveyed by speech and gestures was presented to participants on video rather than in a face-to-face context. This limits the generalizability of their results, since the behavior of individuals may differ in the two situations. Thus, a different context might produce different results. Holler and Beattie (2003) tried to overcome this shortcoming by testing McNeill’s theory of iconic gestures in conversational interaction. The authors investigated the communicational role of iconic gestures and how the representation of semantic information is partitioned between iconic gesture and speech. To capture the range of semantic information conveyed by the two channels, they empirically derived 20 semantic categories (entity, action, relative position, size, shape, shape of a part, and their subcategories). Participants narrated cartoon stories to one of the experimenters. Narrations were filmed and gestural and verbal material was coded through a mathematical scheme, binary for gesture (semantic feature represented or not by the gesture) and tripartite for speech (explicitly represented semantic information, implicitly represented information, no information).

The analysis of the study was structured in such a way that six different speech-gesture combinations of informational values were possible, and it described directly how often gesture and speech interact in a particular kind of pattern. None of the semantic features showed an identical pattern of representation. The distribution of frequencies across the six possible combinations of informational values of each semantic category was homogenous for some of them and heterogeneous for others. The semantic categories were then classified in groups according to how similar their pattern of representation was. The findings suggest that the pattern of how speech and gesture work together to represent semantic information varies considerably from feature to feature. Some were primarily represented by gesture, others by speech. The authors concluded that “the interaction of the gestural and the linguistic systems is more multifaceted than as described by McNeill and that iconic gesture and speech do not necessarily represent different aspects of the same scene. Rather, they only do so with respect to certain semantic features” (Holler & Beattie, 2003, p. 111). They provided two hypotheses as to why some semantic features are represented differently from others. It could be that the semantic features that are similarly represented by speech and gesture bear some kind of common characteristic. This would suggest that gesture and speech are two systems, each designated to represent particular semantic features, and thus, they “operate together in a rather

20

static and fixed manner. […] An alternative hypothesis might be that the gestural and the linguistic systems interact in a much more flexible manner, which is linked to the communicational intent of the speaker” (p. 111). The latter theoretical model would make it not possible to determine which semantic features will be represented by which channel. Gestures may serve different kinds of communicational functions, and the communicational demands of a certain situation may lead to variation in the semantic features represented by iconic gestures. If this latter hypothesis is true, then it would be interesting to study the behavior of a L2 learner while performing the same task in their native language and their second language, in order to understand how certain communicational demands may lead to variation in iconic gestures and add to our understanding of such gestures in relation to second language.

Poggi’s view regarding the semantic features represented by iconic gestures is similar to that of Holler and Beattie. In her paper, she outlines a process through which iconic gestures may be generated. “Creating an iconic gestural noun implies sorting out and miming […] one or a few aspects of the referent that allow the Addressee to restrict his guess about what we are referring to” (Poggi, 2008, p. 52). The features of a referent will be chosen according to the speaker’s goals and communicative resources: the speaker will represent beliefs most distinctive of the referent, possible and easy to be represented by hands. Poggi posits the idea that iconicity in gestures may vary, in terms of the features of gestures resembling the features of the meaning they represent. “The iconicity of a gesture is not an all-or-none matter: there are different levels of iconicity” (Poggi, 2008, p. 55). A criterion to measure iconicity would be to consider the parameters that are iconic in a gesture and resemble aspects of the meaning: the more iconic parameters there are, the more iconic the gesture is. Despite Poggi was referring to codified gestures, those “steadily represented in the mind as lexical items of a gestural lexicon” (p. 48), the criterion to measure iconicity can also apply to iconic gestures in the sense proposed by McNeill.

The concept of iconicity was also later proposed by Perniss and Vigliocco (2014, p. 2) who regard it as “any resemblance between certain properties of linguistic/communicative form (this includes sign or spoken language phonology, sign or spoken language prosody and co-speech gestures) and certain sensori-motor and/or affective properties of corresponding referents”. Traditionally, in sign languages, signs have been classified as transparent signs, translucent signs, obscure signs, and opaque signs depending on how clear, or iconic, their meaning is. Indeed, the iconic form can differ in the extent or degree to which it resembles its referent, and thus can exhibit varying degrees of abstraction. In spoken languages, the authors

21

maintain, co-speech gestures “offer similar opportunities for iconic representation of action affordances and visual features of referents, and therefore, like signs, gestures can exhibit varying degrees of perceptual/motoric iconicity” (Perniss & Vigliocco, 2014, p. 3). The concept of iconicity may become useful when qualitatively analyzing iconic gesture. Considering the parameters that are iconic in a gesture and that resemble aspects of the meaning could be a criterion to study each gesture. If, as posited by Poggi, the features of a referent will be chosen according to the speaker’s goals and communicative resources, the results might provide insights about the speaker’s communicative strategies and in particular what strategies are employed in the speaker’s first and second language to pinpoint potential differences.

The above-mentioned literature shows disagreement on the function of iconic gestures. They might play a role in lexical access, or they might add further information to that expressed through speech and aid communication. The study conducted by Beattie and Goughlan (1999) tested experimentally the first theory inducing a TOT state in participants, providing, however, little evidence that iconic gestures have a functional role in word retrieval. A research conducted by Bettie and Shovelton (1999b) found some evidence in support of McNeill’s theory that iconic gestures convey additional meaning to that expressed through speech. Holler and Beattie (2003) further tested McNeill’s theory and found that the way speech and gesture interact is rather flexible and is linked to the communicational intent of the speaker. The research will employ a qualitative approach to look at iconic gesture and second language, and more specifically to investigate iconic gestures produced by the same speaker while speaking their L1 and L2. This might help us to add to our understanding of such gestures and what we know about their function in L1. The focus of the second research question will be about the functions of iconic gestures and whether variation occurs in the realization of such gestures across participants’ L1 and L2, since, as it has been hypothesized, the communicational demands of a certain situation might lead to variation in the semantic features represented by iconic gestures. Because their form and manner present picturable aspects of the semantic content, it is likely that many iconic gestures will be produced in a storytelling task, and similar repertoires of iconic gestures might be produced across English and Italian narrations. This will allow to compare the behavior of bilingual speakers while performing the same task in their L1 and L2, which might provide some insight on the effect that the communicational demands and linguistic abilities across a speaker’s L1 and L2 might have on gesture production. The concept of iconicity will be employed to qualitatively analyze iconic gesture. Moreover, since previous research has shown that proficiency might affect gesture production, the level of participants in

22

English will be taken into account. Iconic gestures performed when narrating the story in Italian and English will be examined. If such gestures will be found to differ in the two narrations, a third point of the study will seek to examine the nature of such differences, where they may emerge, and to identify any salient features of gestures. If they do differ, it might be that the speaker makes different use of iconic gesture in the two narrations.

23

Methodology

Design of study

In the attempt to motivate naturalistic speech-gesture production, a semi-controlled study was designed. It presents a within-subject, mixed method design, whereby quantitative and qualitative analyses were conducted. A narrative task was chosen, as it is known that gestures are abundant in storytelling (Gullberg et al., 2008). The procedure is the one reminiscent of McNeill (1992). To elicit speech-gesture production in two different languages, Italian and English, participants were exposed to a stimulus video twice. Subsequently, they were asked to narrate the story to a listener, once in English and the other, about a week later, in Italian. To take part in the experiment, participants needed to have a working knowledge of English. Since knowledge is a broad term which is not easily assessable, for the present study knowledge of English equaled the ability to narrate the story of a cartoon to another English-speaking person. To make sure participants met the requirement, they were selected among students at English-speaking universities.

Materials

The stimulus was a short animated cartoon, Sylvester the Cat, on the adventures of Sylvester and Tweety. The episode was “Home, Tweet Home”, approximately 7 minutes long. The plot of the cartoon is about the endless attempts and failures of the cat, Sylvester, that tries to catch the bird, Tweety. The cartoon was shown to participants on a laptop. Moreover, to have an accurate record of the storytelling, a camera was used to audio-video record the narrations.

Participants

13 native speakers of Italian, 8 male and 5 female, whose age ranged 19-32 (M=25.8, SD=3.87), participated in the experiment. Since participants had to retell the story both in their L1 and in English, a fundamental prerequisite to take part in the study was knowledge of the English language. To make sure the requirement was met, participants were selected among university students at Radboud University and at the Donders Institute in Nijmegen, in which English is the language of instruction. Participants had different levels of English that ranged from low intermediate to upper intermediate. None of the speakers were simultaneous bilinguals. All participants learnt English in primary school or later.

24

Procedure

Participants were shown an animated cartoon and were asked to narrate the story to a listener who did not see the cartoon. The experiment was divided in two parts, which took place in two different days, about a week from each other. The first time the participants narrated the story in English, whereas the second in Italian. Different listeners were chosen for the two parts of the experiment. Participants were allowed to watch the cartoon as many times as they liked, until they felt confident in retelling the story. At that point, they were asked to narrate the story in a clear and detailed way to the listener. They could also take notes, which they could review prior to the narration, however, during the narration they were not allowed to look at their notes. Participants were told that they were taking part in a research that investigated the relationship between language proficiency, narrative comprehension, and memory. They were not aware that gestures were of interest in the study, and gestures were not mentioned in the instructions. The participants’ narrations of the stimulus video were videotaped with a camera.

The experiments took place in study rooms at the university or student dorms rather than in a laboratory. The purpose was to try to recreate a communicative situation that would be as naturalistic as possible. Students are quite familiar with such rooms, and therefore it was believed that they would feel at ease in carrying out the experiment there. On the contrary, inviting them into a lab would have negatively affected the naturalness of the gesture-speech production. The rooms had several chairs, all swivel chairs or armchairs, and participants were free to sit wherever they wanted. No instructions were given regarding the sitting position and the positions of the arms and hands, in order to naturalize the production. Moreover, participants watched the cartoon on a laptop rather than on a big screen. These factors contributed to the naturalness of the communicative situation, since, especially university students are well known to watch videos on laptops and sit on swivel chairs.

Establishment of proficiency

In order to establish the English proficiency level of participants an Assistant Professor of English language in the Netherlands provided an assessment of perceived proficiency based on speech production criteria, involving rhythm and pace of production, lexico-semantic accuracy, vocabulary range, and grammatical accuracy.

25

Data sample

The final corpus comprised 26 narrations produced by 13 Italian native speakers. Of these narrations, 13 were in English and 13 in Italian. Proficiency in English varied across participants. Therefore, two categories have been identified to which narrations were attributed to: higher proficiency and lower proficiency. Of the 13 English narrations, 5 were classified as lower proficiency speaker productions and 8 as higher proficiency speaker productions on the basis of their performance when retelling the story to the listener. Overall, more than two thousand gestures were produced in the English and Italian narrations. Therefore, to answer the second research question, a subsidiary corpus was derived from the initial corpus in order to qualitatively analyze iconic gestures. The subsidiary corpus consisted of iconic gestures produced by six participants, of which three were higher proficiency speakers and three were lower proficiency speakers. For each participant, three pairs of iconic gestures were selected, three performed in the English narration and three performed in the Italian narration. For the gestures to be selected, it was important that they were performed while describing the same point of the narrative sequence in both the speaker’s L1 and L2 and that they referred to the same semantic content, in order to allow for a comparison.

Speech transcription

The Italian and English narrations were transcribed by an Italian native speaker, who is also a higher proficiency speaker in English, and checked for accuracy. From each of the Italian narrations, two versions of an English translation were subsequently created. One was a literal, or word by word, translation, where each (or almost) Italian word corresponded to its English counterpart. The other was an idiomatic translation, where the purpose was to transmit the message rather than the literal Italian verbiage. The reason why two types of English translation were made is because they would become useful when interpreting the speech-gesture transcript. The literal translation allows us to accurately associate each Italian word to the corresponding English word, making clear which words were produced during each gesture phase. The idiomatic translation would be helpful in comprehending the meaning of the utterance in cases where the literal English translation would deviate too much from the linear English sentence formation and would make it not possible to understand the meaning of the original Italian utterance. Italian is a null-subject language, whereas English requires an explicit subject in the sentence. Therefore, whenever a null-subject occurred in Italian, a bracketed subject was inserted in the English literal translations. In the idiomatic translations, however,

26

all subjects were normally expressed. All the repetitions, hesitations, restarts, and audible pauses were included in the transcripts. This was done as well for the literal English translations of the Italian narrations, but not for the idiomatic translation.

Gesture transcription

As previously stated, the interest of the present study is in spontaneous gesticulation used in co-occurrence with speech and which is incomplete without speech accompaniment. What was categorized as gestures, therefore, were movements of the hands and arms that co-occurred and were synchronized with speech production. To detect a gesture and to categorize it as such, the focus was on identifying the stroke, which is the meaning bearing phase of a gesture (McNeill, 2007). It is very likely that the stroke co-occurs with a portion that is linguistically articulated and is co-expressive with the gesture. Therefore, gestures produced in the narratives that corresponded to the aforementioned description, where the stroke was identifiable, were counted and later coded adapting the annotative practice employed in the McNeill Lab at the University of Chicago (Duncan, 2005). Gesture phases were identified (preparation, prestroke hold, stroke, stroke hold, poststroke hold, retraction, McNeill, 2007) in order to allow for a comparison of the single segments of the paired gestures in the two languages. In the transcription, the portion of the text that corresponded to the starting and ending points of a gesture was identified with square brackets. To indicate the segment that corresponded to the stroke (the obligatory part in a gesture, McNeill, 2007) the text was bolded. “*” corresponds to audible pauses, whereas “/” to silent pauses. To classify gestures, the distinction made by McNeill (2007) was applied. The categories (or dimensions) gestures can fall under are four: iconic or iconicity; metaphoric or metaphoricity; deictic or deixis; beat or temporal

highlighting.

Speech and gesture transcriptions

In the qualitative analysis a subsample from the corpus of iconic gestures produced by 6 participants (three higher and three lower proficiency speakers) in English and in Italian was selected. This resulted in a representative corpus of 36 instances. Gestures produced by the same speaker were coupled, one coming from the English and one from the Italian narration. Such gestures were representing the same part of the narrations in similar or almost equivalent manner. These particular gestures were selected because interesting and contrasting features were visible in the iconicity of gestures performed by higher and lower proficiency speakers.